Update README.md

README.md

@@ -58,9 +58,19 @@ This dataset can be used for the task of Natural Language Inference (NLI), also
 
 ### Data Splits
 
+The premise counts are distributed as follows:
+
+|     # premises    |# problems|% problems|
+|---------:|------:|------------:|
+|   1    |  192  |     55.5 %     |
+|   2    |  122  |      35.3 %     |
+|   3    |   29  |      8.4 %     |
+|   4    |  2  |      0.6 %     |
+|   5    |  1  |       0.3 %     |
+
 The answer distribution is roughly as follows:
              
-|     Number of problems    |Percentage|Answer|
+|     # problems    |Percentage|Answer|
 |---------:|------:|------------:|
 |   180    |  52%  |      Yes     |
 |   94    |  27%  |      Don't know     |